CLOUD

For decades, high performance computing (HPC) researchers have struggled with low-level programming environments to exploit parallel computers. The Department of Defense's (DoD) High Performance Computing Modernization Program’s (HPCMP) sponsored researchers Dr. Jeremy Kepner, Massachusettes Institute of Technology (MIT) Lincoln Laboratory, Dr. Stan Ahalt, The Ohio State University (OSU) and Ohio Supercomputer Center (OSC), and Dr. John Nehrbass, OSU, have been leading an effort to bring the dominant high-level language of technical computing (MATLAB) into the HPC world. The first stage of this effort has been to bring MatlabMPI (www.ll.mit.edu/MatlabMPI) to OSC and the DoD HPCMP Major Shared Resource Centers (MSRCs). MatlabMPI is an exceedingly small (less than 350 lines) library written entirely in MATLAB, which allows it to run anywhere MATLAB runs (including Windows). As a result, MatlabMPI has become the leading parallel MATLAB solution with hundreds of users worldwide. In spite of its small size, MatlabMPI has achieved speedups as great as a 300 speedup factor on 304 processors. OSC and the HPCMP MSRCs are the first large HPC centers to deploy this technology and were selected because of their large MATLAB user base and the on-site training expertise OSU has built up in the area of parallel MATLAB.